| { | |
| "architectures": [ | |
| "ArgonneModel" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "block_size": 4096, | |
| "eos_token_id": 151645, | |
| "hidden_dropout": 0.0, | |
| "hidden_size": 4080, | |
| "intermediate_size": 11008, | |
| "max_position_embeddings": 4096, | |
| "mlp_bias": false, | |
| "model_type": "argonne2", | |
| "n_embd": 4080, | |
| "n_head": 24, | |
| "n_layer": 24, | |
| "num_attention_heads": 24, | |
| "num_hidden_layers": 24, | |
| "num_key_value_heads": 8, | |
| "pad_token_id": 151643, | |
| "rms_norm_eps": 1e-06, | |
| "rope_theta": 10000.0, | |
| "sliding_window": null, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.44.0", | |
| "use_flash_attention": true, | |
| "use_gradient_checkpointing": false, | |
| "vocab_size": 151665 | |
| } | |